Grad student seeking info from pizzeria operators

[rhuff]

Charles:

It was a meaningless survey with the sample size and methodology. That is what I was questioning in the first place. I just expected more from a graduate project. I could see this as an attempt by an undergrad, but at the graduate level it is astonishing, especially if he really does go to that institution.

It may be meaningless to you, but its useful to me. As I said, its a survey, not a scientific study. Small sample sizes are used all the time by people and companies to help filter things out, or show promise that warrants additional, more substantial study. If I was doing an FDA drug effectiveness test, then it would be an entirely different story. The practical fact is that I don’t have the resources to do a more controlled study where I collect responses from a sampling that represents the overall population.

As an example, one very useful finding is that some pizza operators spend over $3,000/month on coupon distribution. It also seems to show that there may be a deep disparity in the use of coupons between different operators. If the results were more uniform, then I wouldn’t spend much time trying to understand coupon consumption habits. But given the deep disparity, I will look into this further.

 

[Charles]

If your methodology and sampling are not correct, the results are meaningless. What if you just happened to have 14 people that all came in with a low standard deviation? Would you then claim, “Oh, I guess there is not much disparity.” That, too, would be erroneous. You simply cannot derive anything from a self-selected survey of such low size. If you are a graduate student and I hope have taken statistics, you should know this. Maybe you just want to get that paper in for a grade and be done with it. But you cannot learn anything meaningful from the data you collected because you simply have too high a margin of error.

Bad information is worse than no information.

 

[rhuff]

Yes, I did take statistics. Thank you. What you seem to not understand is that it is acceptable in some circumstances to use non-random sampling. It depends entirely in the context that the results are used. Its done all the time.

 

[Charles]

http://imgs.xkcd.com/comics/correlation.png

 

[system]

Could I get a copy of the survey please?

Thanks

 

[NicksPizza]

rhuff:

NicksPizza:

Most of your standard deviations are insane looking?? Your SD is larger than your average in three questions??? Something is amiss, or the small sampling is really goobering your numbers.

I saw that too. The numbers were all over the place, so the standard deviation doesn’t have much value. Its interesting that there is so much variety in how pizzeria operators use coupons. Here is the detail for the coupon answer:

Please understand our (MY?) protectiveness of the industry. There are soooo many lousy “research” studies floating around claiming all sorts of things supposedly relavent to food service, marketing, small businesses. I have a personal weakness in my personality and patience when it comes to research and statistical analysis. I know just enough to be annoyed at weak designs and faulty analysis. It has been far too many years since my Masters thesis utilizing a point-biserial MANOVA to analyze combinations of data types from Lickert Scales to ratio data to binary info to percentile scores (1993?). I had sort of weak analysis due to limited sample of complete data sets. I know the hazards of small sampling.

The value of the stadard deviations you have is to tell you that your varaiance is way too wide for any meaningful statistical analysis, and that the arithmetic mean is the truly useless metric. You will probably have major problems even managing the data using logarythmic or polynomial trendlines to “normalize” the data.

Matched sampling and other sampling methods make sense in lots of situations. If this is just a “fluffy” little exercise, then I stand apologetic. If you are trying to make real and utilized/published assertions about coupon use in the pizzeria industry, you’ll hit all sorts of knee-jerk reactions in this community. So many different strong opinions . . . and there is an economic strife in effect right now for many of us.

In the end, it is your project and your class. We will genrally understand that while we will rant on . . . you will do what the class demands and accepts.